Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
74
Data Hiding in Binary Images Using Orthogonal Embedding - A High
Capacity Approach
Bharti Gawali
1,
Dr.Babasaheb Ambedkar Marathwada University,
Aurangabad, India
E-mail: [email protected]
Rupali Kasar2,
Dr.Babasaheb Ambedkar Marathwada University,
Aurangabad, India
E-mail: [email protected]
Abstract
The growth of high speed computer networks and the Internet, in particular, has increased the ease of Information
Communication. In comparison with Analog media, Digital media offers several distinct advantages such as high quality, easy
editing, high fidelity copying, compression etc. But this type advancement in the field of data communication in other sense
has hiked the fear of getting the data snooped at the time of sending it from the sender to the receiver. Information Security is
becoming an inseparable part of Data Communication. In order to address this Information Security, Digital Watermarking
plays an important role. Watermarking Techniques are used to hide a small amount of data in such a way that no one apart
from the sender and intended recipient even realizes there is a hidden message. This paper proposed a high capacity data hiding
approach for binary images in morphological transform domain for authentication purpose so that the image will look
unchanged to human visual systems.
Keywords: Data hiding, digital watermarking, authentication, security, authentication, binary image.
I. Introduction
Digital watermarking is the process of embedding information into a digital signal. The signal may be audio, pictures or video.
If the signal is copied, then the information is also carried in the copy [1].A significant number of data hiding techniques have
been reported in recent years in order to create robust digital watermarks. Among all the existing techniques for digital color
and gray scale images, not all of them can be directly applied to binary text images. Digital watermarking techniques have been
proposed for ownership protection, copy control, fingerprinting, annotation and authentication of digital media [2]. Recently
authentication of digital documents has found wide applications in handwritten signatures, digital books, business documents,
personal documents, maps, engineering drawings, and so on [3]–[4]. Most data-hiding techniques for binary images have
focused on spatial domains, for example, choosing data-hiding locations by employing pairs of contour edge patterns [5], edge
pixels and defining visual quality-preserving rules [6], [7]. However, the capacities of the existing algorithms are not large
enough, especially for small images.
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
75
Figure.1.Illustration of changes in coefficient due to flipping pixels [(a)and(b)]and the “edge shifting ”phenomenon [(c) and (d)].
A watermarking scheme is usually consist of three distinct steps,
Embedding,
Attack and
Detection.
Authentication process ensures that the given set of data comes from a legitimate sender and content integrity is preserved [8].
Watermarking can be further classified into hard authentication and soft authentication [9]. Hard authentication is the
validation of content that does not allow any modifications. That means a single bit change in the test image will trigger an
alarm that indicates the content is unauthentic. “Hard” image authentication is highly sensitive and dependent on the exact
value of image pixels, whereas “Soft” image authentication is sensitive just to content modification and serious image quality
tampering.
On the other hand, editing an image becomes easier with the powerful image editing tools and digital cameras. Authentication
to detect tampering and forgery is thus of primary concern. This work focuses on hard authenticator watermark-based
authentication. The problem of data hiding for binary images in morphological transform domain is concentrated. The
morphological binary wave transform [10] can be used to track the transitions in binary images by utilizing the detail
coefficients as a location map to determine the data hiding locations. Embedding data using real-valued coefficient requires
more memory space. The Morphological binary wavelet transform [10] can be used to track the transitions in binary images by
using the detail coefficients. This coefficient is used as a location map to determine data hiding locations but the problem lies
with the fact that once a pixel is flipped ,the horizontal, vertical and diagonal detail coefficients also changes. The major
advantages of the proposed scheme lie in its larger capacity compared with previous schemes, better visual quality (e.g.,
compared with [5]–[9]) and lower computational complexity. In addition, unlike [4], our present scheme does not suffer the
capacity decrease and computational load increase in order to incorporate the cryptographic signature for authentication.
The paper is divided into seven sections. Section I, deals with the introduction of digital watermarking. Section II, provides a
brief review on existing 1-D morphological binary wavelet transform, our proposed extension to 2-D morphological binary
wavelet transform is presented in Section III, Section IV describes the Orthogonal Embedding method, Experimental results
and data hiding effects are presented In Section V and VI ,Section VII concludes the paper.
II. 1-D Morphological Binary Wavelet Transform
An interlaced transform is used to identify the embeddable locations due to the fact that some transition information is lost
during the computation of a single transform and there is a need to keep track of transitions between two and three pixels for
binary images data hiding. Here an image based on 2x2 pixel blocks is processed and two different processing cases are
combined that the flippability conditions of one are not affected by flipping the candidates of another for data embedding are
called “Orthogonal Embedding”. This addresses the problem of the capacity decrease due to the un-embeddability of the block
boundaries. As a result, significant gains in capacity can be achieved, which also improves the efficiency of utilizing the
flippable pixels. “Exclusive OR (XOR)” operation addresses the quantization error occurred in a DCT-based approach. The
major advantages of the proposed scheme lie in its larger capacity (e.g., compared with [5], [11], [12]), better visual quality
(e.g., compared with [5]–[7]) and lower computational complexity (e.g., compared with [5], [4]). In addition, unlike [7], our
present scheme does not suffer the capacity decrease and computational load increase in order to incorporate the cryptographic
signature for authentication.
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
76
This section, reviews with signal analysis and synthesis and 1-D signal decomposition similar to [14].
Based on this, decomposition scheme to 2-D signal is extended in this paper and subsequently proposed an interlaced transform
for the data-hiding application.
Consider a family of signal space V𝒍 and detail space W𝒍 at level 𝒍. The 1-D wavelet decomposition scheme comprises of one
signal analysis operator ψ𝒍 + and one detail analysis operator ϖ 𝒍
+. In addition, it also consists of one signal synthesis operator ψ𝒍
⎯and one detail synthesis operator ϖ 𝒍⎯.Here “+” indicates that the corresponding operator maps a signal to a higher level,
towards the direction of reducing the information. Whereas “_”indicates that the operator maps a signal to a lower level,
towards the direction of restoring the information. The signal analysis operator ψ𝒍 + maps a signal, S𝒍 ∈ V𝒍
from level 𝒍 to 𝒍+1 for the scaled signal S𝒍+1 = ψ𝒍 + (S𝒍), S𝒍+1 ∈ V𝒍 +1 The detail analysis operator ϖ 𝒍+
V𝒍 ⟶ W𝒍+1 maps a
signal from level 𝒍 to 𝒍+1 for the detail signal d𝒍+1 = ϖ 𝒍+
(S𝒍), d𝒍+1∈ w𝒍+1 On the other hand, the signal synthesis operator
ψ𝒍 - : V𝒍+1 ⟶ V𝒍 maps a signal from level 𝒍+1 to back to level 𝒍 to obtain an approximation of S𝒍 , to S𝒍 = ψ 𝒍
-(S𝒍+1 ) The detail
synthesis operator ϖ 𝒍⁻ (d𝒍+1)maps a detail signal back to level 𝒍 so as to obtain the detail signal e𝒍 = ϖ𝒍 -( d𝒍+1) The signal at level
𝒍 is reconstructed by
S𝒍 = ψ𝒍 –
(S𝒍+1 ) + ϖ 𝒍 – (d𝒍+1 ) (1)
Perfect reconstruction of the original signal is possible if the operators satisfy
ψ𝒍 – ψ𝒍
+ = ϖ 𝒍
- ϖ 𝒍+ = I (2) ψ𝒍
–ϖ 𝒍 +
= ϖ 𝒍 - ψ𝒍 + = 0 (3)
ψ𝒍 – ψ𝒍
+ + ϖ 𝒍
- ϖ 𝒍+= I (4)
Where “I” and “0” represent the identity and zero operators, respectively. Note that that {ψ𝒍–,ψ𝒍
+,ϖ𝒍
-,ϖ𝒍
+} is a set of
biorthogonal filter operators if the conditions in (2)–(4) are satisfied. For the morphological analysis and synthesis scheme the
bi-orthogonality is formulated in the operator terms. Let “ ” denote the “Exclusive OR (XOR)” operation. The signals at
level 𝒍+1 by applying the analysis operators for a 1-D signal are given by
s(i) = ψ+(s)(i) = s(2i+1) (5)
d(i) = ϖ+ (s)(i) = s(2i) s(2i+1) (6)
s(i) & d(i) are the coarse and detail signals obtained at level 𝒍+1,respectively. i denotes the index of the signal at level 𝒍+1 and
i = 0,1,2,…..,N-1 for a 1-D signal of size N . The detail signal d(i) contains 1’s at each transition from 0 to 1 or vice versa in
signal. The synthesis operators are given by
ψ –(s) s(2i+1) = ψ
–(s)(2i) = s(i) (7) ϖ
–(s) s(2i+1) = 0 and ϖ –(s) (2i) = d(i) (8)
The signal at level 𝒍 is reconstructed by
s(2i) =ψ –(s)(2i) ϖ
–(s) (2i) = s(i) d(i) (9)
s(2i+1) =ψ –(s)(2i+1) ϖ
–(s)(2i+1)= s(i) (10)
III. Interlaced Morphological Binary Wavelet Transform
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
77
The 1-D wavelet decomposition scheme [14] is extended to a 2-D signal by defining a non-separable 2-D transform. Let i and j
represent the indices of the signal at level l +1, where i=0,1,2,…M-1 and j = 0,1,2,…N-1 for a 2-D coarse signal of size M x
N .Designation of the samples in a 2 x 2 block is shown in Fig. 2.where s (2i,2j) denotes the signal (“0” or “1”) located at row
and column at level .To define a 2-D transform, one sample in a 2 x2 block is sub-sampled as the coarse signal.
Figure 2. Designation of the samples in a 2 x 2 block
The horizontal, vertical and diagonal detail signals are derived from the differencebetween the sub-sampled sample and its
vertical, horizontal and diagonal (including diagonal and anti-diagonal) neighbors. The resultant transformed signal remains
binary while the coarse and detail signals will each be 1/4 the size of the original signal. Let the operators for the coarse signal,
horizontal, vertical, and diagonal detail signals be, ψee
, ϖh ee
, ϖvee
and ϖdee
, respectively, where the superscript “ ee ” denotes
“even-even”. The obtained transform is named the even-even transform since it is operated on a 2 x 2 block starting from the
even-even coordinates. The coarse signal, vertical, horizontal and diagonal detail signals at level l +1 are obtained by applying
the analysis operators to obtain.
See(i,j) = ψee+ (S) (i,j) = S (2i+1,2j+1) (11)
vee(i,j)=ϖhee+ (S) (i,j) = S (2i+1,2j) S(2i+1,2j+1) (12)
hee(i ,j ) = ϖ ee+ (S) (i, j) = S (2i,2j+1) S(2i+1,2j+1) (13)
dee(i ,j )= ϖd ee+ (S) (i, j) = S (2i,2j) S(2i+1,2j) S(2i ,2j+1)
S (2i+1,2j+1) (14)
Where See(i,j) ∈ V𝒍+1 and { vee(i,j), hee(i ,j ) , dee(i ,j )} ∈ W𝒍+1. the synthesis operators of a 2-D wavelet transform are given
by
ee ee
ee
ee
ee
Finally, the signal at level l+1 can be reconstructed by
S(2i,2j)=
ee (s)(2i,2j) ϖ𝑑 ee-(S) (2i,2j)
=See(i,j) vee(i,j) hee(i,j) dee(i,j) (16)
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
78
S(2i,2j+1)=
ee (s)(2i,2j+1) ϖh ee-(S) (2i,2j+1)
= See(i,j)
hee(i ,j) (17)
S(2i+1,2j)=
ee (s)(2i+1,2j) ϖv ee-(S) (2i+1,2j)
= See(i,j)
vee(i ,j) (18)
S(2i+1,2j+1)=
ee (s)(2i+1,2j+1) ϖ ee-(S) (2i+1,2j+1)
= See(i,j) (19)
Since the coarse signals obtained from (11) are at the odd-odd locations, the transitions from odd-odd to other coordinates in
the 2 x 2 block are assessed from the detail signals obtained by (12)–(14). To obtain the transition from odd-even, even-odd
and even-even coordinates complementary wavelet transforms operating on the 2 x 2 blocks is designed which start from the
even-odd, odd-even and odd-odd coordinates of the signal. Based on the absolute coordinates in the top left position shown in
Figure.3. Each 2 x 2 block in a 2-D image is classified as an even-even block (EEB),or an even -odd block (EOB), or an odd-
even block (OEB), or an odd-odd block (OOB), which is given by
Figure 3. Four different 2 x 2 blocks in 3 x 3 blocks of binary image.
Bp(i,j)∈ EEBs for (mod(x,2)=0)^ (mod(y,2)=0)
EOBs for (mod(x,2)=0)^ (mod(y,2)=1)
OEBs for (mod(x,2)=1)^ (mod(y,2)=0)
OOBs for (mod(x,2)=0)^ (mod(y,2)=1) (20)
Where (x,y) denotes xth row and yth column of an image ,(i ,j) denotes the index of the current 2x2 block mod() denotes
a modulo operation and “ ^ “ represents logical “AND”operation. Hence, three additional transforms, i.e., even-odd, odd-even
and odd-odd, can be defined. These transforms, together with the even-even transform, are collectively called interlaced
morphological binary wavelet transform (IMBWT).The operators for the even-odd transforms are , the signals obtained by
applying the analysis operators for the even-odd transform are given by
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
79
Figure. 4. Main processing block and its subsidiary block
Seo(i,j) = ψeo+ (S) (i,j) = S (2i+1,2j+2) (21)
veo(i,j) = ϖv eo+ (S) (i,j) = S (2i+1,2j+1) S(2i+1,2j+2) (22)
heo(i,j) = ϖheo+ (S) (i, j) = S (2i,2j+2) S(2i+1,2j+2) (23)
deo(i ,j ) = ϖd eo+ (S) (i, j) = S (2i,2j+1) S(2i+1,2j+1)
S(2i ,2j+2) S (2i+1,2j+2) (24)
The odd-even and odd-odd transforms can be defined in the same way. For simplicity, we use, sk ,vk ,hk , and dk ,
k ∈ {ee,eo,oe,oo} to represent the signals obtained from different transforms. There are four single processing cases (SPCs):
even-even, even-odd, odd-even and odd-odd that are determinant on the main processing blocks to be EEBs, EOBs, OEBs and
OOBs. Consider the even-even processing case where the main processing blocks are EEBs and EOBs, OEBs and OOBs which
are interlaced with the EEBs are shown as the subsidiary blocks in Fig. 4. In Applying the IMBWT for data hiding in binary
images, the processing of images is always based on 2 x 2 blocks (i.e., main processing blocks). However, the flippability of a
coarse signal is determined by considering 3 x 3 blocks, which consist of both the main processing blocks and the subsidiary
blocks.
III.I Single Processing
As flipping an edge pixel in binary images is equivalent to shifting the edge location horizontally one pixel and vertically one
pixel. A horizontal edge exists if there is a transition between two neighboring pixels vertically and a vertical edge exists if
there is a transition between two neighboring pixels horizontally. To define the flippability condition for a coarse signal, we
actually consider the 3 x 3 block in such a way that the shifted edges can be tracked for the convenience of blind watermark
extraction. The flippability condition (or cross condition) for a SPC is defined as follows: a coarse signal (in a main processing
block) is flippable if both horizontal and vertical edges exist. Specifically, consider the even-even processing case, a horizontal
edge exists if either hee or hoe equals to 1 and a
vertical edge exists if either or equals to 1. The flippability condition f ee(i,j) is given by
f ee(i,j)= (vee(i,j) veo(i,j))) ^ (hee(i ,j )
heo(i ,j)) (25)
f ee(i,j)=1. (26)
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
80
Let us now consider a 3 x 3 block of an input image s. The number of transitions from the current candidate pixel (the center
pixel) to its 4-neighbors in the horizontal and vertical directions is represented by Nh and Nv, respectively, which are calculated
from the center pixel to its 4-neighbors. Assume the center pixel S (2i+1, 2j+1) is [see Fig. 3(a)], and Nh and Nv are given by
Nh = ∑ s(2i+1,2j+k) s(2i+1,2j+1+k) (27)
Nv = ∑ s(2i+k,2j+1)
s(2i+1+k,2j+1) (28)
By satisfying the cross condition that fee(i,j)=1 and Nv do not change when the center pixel is flipped. Hence, the 4-
connectivity from the center pixel to its 4-neighbors is preserved. Further, the center pixel has a white 4-neighbor pixel in both
horizontal and vertical directions. Similarly, the flippability conditions for the odd-odd, even-odd, and odd-even processing
cases is calculated. In summary, the flippability conditions help identify all the candidate pixels for which the pixels directly
above and below the candidate pixel have different colors (“0” and “1”) and the pixels immediately to the left and right of the
candidate pixel have different colors. In addition, in applying these flippability conditions for data hiding, we require that the
two chosen candidate pixels should not be 4-adjacent, i.e., horizontally or vertically adjacent to each other to avoid poor visual
quality. It can be seen that the flippability of the center pixel in each 3 x 3 block is independent of the center pixel value.
IV. High Capacity data Hiding Using Orthogonal Embedding
The process of orthogonal embedding consists of following processes.
IV.I Authentication Watermark Generation, Embedding and Verification
Firstly it is necessary to apply the flippability.
IV.I.I Flippability decision
In applying the IMBWT for data hiding in binary images, the processing of images is always based on 2X2 blocks (i.e., main
processing blocks). The flippability conditions of a coarse signal, is determined by considering 3X3 blocks, which consist of
both the main processing blocks and the subsidiary blocks
IV.I.II Embeddability
In order to generate the hard authenticator watermark, the key issue is how to locate the “embeddable” pixels given the
watermarked image. The “embeddability” of a block depends on the “flippability” of the determined pixels in the block.
IV.I.III Watermark embedding
The watermark embedding process is summarized as follows.
1) Partition the image into equal size square blocks; note that the block size does not need to be square.
2) Determine the flippability of the determined pixels based on the “Flippability Criterion”.
3) Once a pixel is identified as “flippable”, the block is marked as “embeddable”. The current “flippable” pixel is
Identified as the “embeddable” pixel, i.e. “embeddable” location of the block.
4) Proceed to the next block.
5) Repeat steps 2 to 4 until all the blocks are processed.
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
81
6) Embed the watermark in the “embeddable” blocks by flipping the “embeddable” pixels.
IV.I.IV Authentication process
To address the security concerns, method proposes to generate the Authenticator watermark in fig. by encrypting the XORed
value of the replicated hash value of the binary images and the payload watermark.
IV.I.V Verification process
Extract the watermark based on flippability & embeddable pixel criterion. Thus the tampering can be detected.
The proposal of processing an image in 2X 2 blocks not only prevents the computational load from getting high for blockbased
approach with large block size but also improves the efficiency of utilization of flippable pixels., the flippability conditions
help identify all the candidate pixels for which the pixels directly above and below the candidate pixel have different colors
(“0” and “1”) and the pixels immediately to the left and right of the candidate pixel have different colors. In addition, in
applying these flippability conditions for data hiding, we require that the two chosen candidate pixels should not be 4-adjacent,
i.e., horizontally or vertically adjacent to each other to avoid poor visual quality. It can be seen that the flippability of the center
pixel in each 3X3 block is independent of the center pixel value.
IV.II Double Processing
The capacity of the proposed scheme is determined by the number of pixels that satisfy the flippability condition. This capacity
can be increased significantly by combining the two single processing cases, namely, Double Processing Case (DPC). Based on
the observations of (27) and (28), the flippability condition of the even-even processing case (i.e. fee(i,j)) is not affected by
flipping the candidates in the odd-odd processing case (e.g., S (2i, 2j) and vice versa, as illustrated in Fig. 5. This
“nonintersection” property of the two processing cases renders the processing of the even-even and odd-odd cases can be done
together. The same applies to the even-odd and odd-even processing cases. For convenience, we call the two combined
processing cases as a “Pair Case”.
Figure.5. Designated 2 x 2 blocks.
As evidenced from the increase in the number of candidate pixels, i.e., from [1/4 x M x N] to [1/2 x M x N] for an image of
size, the capacity can be approximately doubled by combining the two processing cases. This idea is motivated by the
quantization index modulation based data-hiding method [18]. The even-even and odd-odd processing cases can be viewed as
two orthogonal sets of embedders indexed at the 2 x 2 blocks starting from even-even and odd-odd coordinates, denoted as Qee
and Qoo , respectively.
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
82
In this paper, the embedders are an ensemble of the embedding functions of a “Pair Case”, e.g., the embedding functions of
even-even and odd-odd cases.
In this approach. We define a designated 2 x 2 block Db(i,j) as the 2 x 2 block that contains the two candidate pixels for a pair
case, which is shown in Fig. 6. The designated 2 x2 block can be chosen from one of the two processing blocks of a pair case,
but only one candidate pixel is chosen to hide data in each. Db(i,j) Hence, the maximum capacity Cc is upper bounded by Cc <=
[1/4 x M x N] To achieve higher security, the choice of an Embedder Qk in each Db(i,j) using DPC can depend on a random
key s where s ∈ {0,1}.. For example, for the mth
Db(i,j) block, we choose ee when s (m)=1 and choose oo when s
(m)=0.. By choosing ee. we check fee(i,j) first, mark the current candidate pixel as embeddable if fee(i,j) =1 and proceed to
the next Db(i,j). Otherwise foo(i,j),will be checked (i.e., when fee(i,j) ≠ 1). Similarly, foo(i,j) is checked first by choosing oo..
From the above discussion, it is noticed that there may be two candidate pixels being flipped in some 2 x 2 blocks (not Db(i,j)),
e.g., the OOBs in Fig.6.
V. Experimental Results
To show the capacity increase by employing DPC compared with that of SPC, 100 images of a variety of sizes were used.
These images are of different types (e.g., cartoons, sketch, drawing, handwritten and text in different languages) and various
dimensions (e.g., 132x132, 128 x 128,232x202, 720x756, and 450x524). The Capacities of the watermarked image, which is
defined in Table 1. The increase in capacity represented in percentage from SPC to DPC.
The capacity mainly depends on the contents of the images, which varies for different types of images. The richer the contents
of the images (more edges exist), the larger the capacity. For images of same size and similar contents, in general, text or
cartoon images which consist of rich corners and thin strokes may easily cause larger distortion. This is due to the facts that
few good patterns exist in corners and thin strokes, hence flipping pixels can be easily noticed. Following figures indicates the
data hiding effects.
Authentication Results
Fig.6. (a) Original Image Of Size
132x132.Max. Data Hiding Capacity for SPC=104,DPC=256, (b)Watermarked image with 80 bits embedded
by employing SPC
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
83
(c)Difference Image with SPC flipped pixels (d)Watermarked image with 40 bits
embedded by employing DPC
(e)Difference Image with DPC flipped pixels
VI. Conclusion
This paper, present a high-capacity approach for data-hiding scheme for binary images based on the interlaced morphological
binary wavelet transforms. The relationship between the coefficients obtained from different transforms is utilized to identify
the suitable locations for watermark embedding such that blind watermark extraction can be achieved. Two processing cases
that are not intersected with each other are employed for orthogonal embedding in such a way that not only can the capacity be
significantly increased, but the visual distortion can also be minimized. Comparative results show that the present scheme is
quite superior in being able to attain larger capacity while maintaining acceptable visual distortion and low computational cost.
References
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
84
[1] I. J. Cox, M. L. Miller, and J. A. Bloom, San Mateo, CA: Morgan Kaufmann,( 2001),
Digital Watermarking.
[2] B. Furht and D. Kirovski, B. Furht and D. Kirovski, Eds. Boca Raton, FL: CRC(2005),
Multimedia Security Handbook.
[3] Y. Liu, J. Mant, E. Wong, and S. H. Low (1999), “Marking and detection of text
documents using transform-domain techniques,” in Proc. SPIE, San Jose, CA, vol. 3657,
pp. 317–328.
[4] Q. Mei, E. K. Wong, and N. Memon (2001), “Data hiding in binary text document,”
in Proc. SPIE, vol. 4314, pp. 369–375.
[5] Y. C. Tseng and H.-K. Pan, “Data hiding in 2-color images (Jul. 2002),” IEEE
Trans. Comput., vol. 51, no. 7, pp. 873–878,.
[6] H. Lu, A. C. Kot, and Y. Q. Shi (Feb. 2004), “Distance-reciprocal distortion measure
for binary document images,” IEEE Signal Process. Lett. vol. 11,no. 2, pp. 228–231.
[7] H. Yang and A. C. Kot (Apr. 2007), “Pattern-based date hiding for binary images
authentication by connectivity-preserving,” IEEE Trans. Multimedia, vol. 9, no. 3,
pp. 475–486.
[8] B. B. Zhu, M. D. Swanson, and A. H. Tewfik(Mar.2004) , “When seeing isn’t believing,”
IEEE Signal Process. Mag, pp. 40–49.
[9] H. Lu, X. Shi, Y. Q. Shi, A. C. Kot, and L. Chen (2002), “Watermark embedding in
DC components of DCT for binary images,” in Proc. IEEE Workshop on Multimedia
Signal Processing, Dec. 9–11, pp. 300–303.
[10] H. J. A. M. Heijmans and J. Goutsias (Nov. 2000), “Nonlinear multiresolution signal
decomposition schemes-part I: Morphological wavelets,” IEEE Trans. Image Process.,
vol. 9, no. 11, pp. 1897–1913.
[11] M. Wu and B. Liu (Aug. 2004), “Data hiding in binary images for authentication
and annotation,” IEEE Trans. Multimedia, vol. 6, no. 4, pp. 528–538.
.
[12] H. Y. Kim and R. L. de Queiroz (2004) , “Alteration-locating authentication
watermarking for binary images,” in Proc. Int. Workshop Digital
Watermarking, pp. 125–136.
[13] H. J. A. M. Heijmans and J. Goutsias (Nov. 2000), “Nonlinear multiresolution signal
decomposition schemes-part II: Morphological wavelets,” IEEE Trans.
Image Process., vol. 9, no. 11, pp. 1897–1913.
[14] “Digital Watermarking” available at
http://en.wikipedia.org/wiki/Digital_watermarking
[15] “Fundamentals of Wavelets” available at
Computer Engineering and Intelligent Systems www.iiste.org ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online) Vol 2, No.3
85
http://documents.wolfram.com/applications/wavelet/index2.html
VII. Observation Table I
Comparison of data hiding Capacities for SPC and DPC
Figure.7. Graph- Capacity Increase Using SPC and DPC
The above fig.7 shows the graph of image index versus Data hiding capacities. It consists of 5 different class images.
Observations shows that capacity has increased significantly using DPC versus SPC, e.g., increase from approximately 56% to
69% for the sample five test images.
Image SPC
capacity
DPC
Capacity
Increase in
Capacity (In characters) Embed data in
characters Lily.bmp 13 32 19
Text.bmp 0 15 15
Drawing.bmp 3 9 6
Tiger.bmp 159 351 192
Sparkle.bmp 2 9 7
Sketch.bmp 13 30 17
Letter.bmp 4 8 4
Graphics.bmp 6 15 9
Nano.bmp 90 191 101
Hill.bmp 56 150 94
Rangoli.bmp 17 43 26
Cartoon.bmp 189 429 240